Spaces:
Sleeping
Sleeping
Commit History
-
Update product database and vector store management with new features and optimizations
- Updated `init_state.json` and `products.db` with new file versions. - Added ONNX model files for CLIP text and vision. - Enhanced `ProductDatabase` class with methods for deleting and updating products. - Improved `VectorStoreManager` to handle existing embeddings and added collection deletion functionality. - Introduced performance optimizations in the reranking service and vector store interactions. - Added performance testing script using Locust. - Created OpenAPI documentation for the API endpoints. - Removed outdated demo test script and added test cases for product CRUD operations.fe2866c
-
Initialize SDK and fix vector store issues
- Add Python SDK with client, models, and examples - Add multi-vector-store support (ChromaDB, Pinecone, Qdrant, Weaviate, PgVector) - Fix image embedding generation with robust retry logic - Fix ChromaDB metadata issue (remove text embeddings from metadata) - Fix search parameter mismatch (query_embedding -> query_vector) - Fix batch processing to handle individual product failures - Add comprehensive debugging and error logging - Add auth, database adapters, and config management - Add Docker support with docker-compose - Update .gitignore with project-specific entries - Remove cached files from trackinga475b1f